Preferential sites of early DNA cleavage in apoptosis and the pathway of nuclear damage.

We have tested the specific hypothesis that the pathway of nuclear collapse in apoptosis is governed by the early attack on active chromatin at spatially restricted nuclear sites. Cell death in PC12 pheochromocytoma cells deprived of serum growth factors, in HL-60 leukemic cells treated with inhibitors of protein or RNA biosynthesis, and in U937 histiocytic lymphoma cells exposed to the cytokine tumor necrosis factor alpha showed a common mechanism in the targeting of DNA for degradation. An incorporation assay with labeled nucleotide revealed an early selective nicking in peripheral nuclear chromatin with concomitant diminution in the amount of immunoreactive lamin B protein. This was followed by a phase of more extensive cleavages, continued nuclear protein loss, chromatin collapse, and fragmentation of nuclei. The spatial restriction of early cleavages is similar to the nicking obtained by the application of exogenous DNase I to fixed nuclei of normal cells and to that obtained in the activation of the endogenous endonuclease of liver nuclei by Ca2+. These similarities suggest that, in apoptosis, activation of an endonuclease preferentially recognizing a specific chromatin configuration, such as that of active (DNase I-sensitive) genes, underlies the early spatial demarcation of cleavages.

Repositioning of human interphase chromosomes by nucleolar dynamics in the reverse transformation of HT1080 fibrosarcoma cells.

An experimental system which should be valuable for studying the role of spatial positioning of the nuclear genome in human cell function has been developed. Reverse transformation of the malignant HT1080 fibrosarcoma cell line upon treatment with 8-chloro-cAMP results in growth inhibition, cytoskeletal reorganization, changes in nuclear shape and chromatin accessibility, and formation of prominent nucleoli. Fluorescent in situ hybridization was used to study DNA positioning during nuclear remodeling. Morphometric analysis of the hybridization sites for both repetitive sequences and "painting probes" for whole chromosomes indicated dispersal of acrocentric chromosomes in untreated cells and a highly organized central location of these ribosome gene-containing chromosomes in association with one or a few large nucleoli in nondividing treated cells. The results suggest that there was a directed movement of interphase chromosomes during a response which normalized a malignant cell line. These large-scale repositionings may serve two functions in restoring a normal transcriptional setup to the nucleus. First, ribosome genes are placed in the nucleolus, their transcriptional suborganelle. Second, nucleolar anchorings together with additional perinucleolar centromeric associations orient the domain shapes of entire chromosomes, installing gene-rich chromosomal regions into pockets of (accessible) DNAse I-sensitive chromatin populated by spliceosomes.

The farnesyl protein transferase inhibitor BZA-5B blocks farnesylation of nuclear lamins and p21ras but does not affect their function or localization.

BZA-5B is a peptidomimetic inhibitor of protein farnesylation in mammalian cells. We have examined the specificity of this compound toward inhibition of farnesylation of p21ras and the nuclear lamin proteins, prelamin A and lamin B. We have also used the Raney nickel cleavage technique in conjunction with radio-gas liquid chromatography to assess the ability of this compound to block total protein farnesylation. These studies show that BZA-5B blocks farnesylation of the lamin proteins with an IC50 comparable to that seen for p21ras. At a concentration in excess of 25 microM, BZA-5B inhibits all protein farnesylation in CHO-K1 cells below the limits of detection. Furthermore, we found that after a 2-day exposure to high concentrations of BZA-5B, CHO-K1 cell lines exhibit no loss in sensitivity to inhibition of prenylation by this compound. Yet, despite the potent and general inhibition of protein farnesylation, BZA-5B does not interfere with a variety of cellular functions expected to be farnesylation dependent, including cell growth and viability, assembly of the nuclear lamina, membrane association of p21ras, and p21ras-dependent differentiation of PC-12 cells in response to treatment with nerve growth factor. The maintenance of farnesylation-dependent events in the presence of BZA-5B stands in marked contrast to the inhibition of the oncogenic ras-mediated transformed phenotype that has been observed with this compound and other farnesyl protein transferase inhibitors. This specificity for inhibition of ras transformation by BZA-5B is quite encouraging to its eventual development as an antimalignancy pharmaceutical.

Reverse transformation and genome exposure in the C6 glial tumor cell line.

Reexpression of growth control and differentiation in response to physiological inducers can be demonstrated in some malignant cell lines, showing that they are not irreversibly transformed. This switch in phenotype is likely to reflect a changing pattern of gene expression, but it has not been known whether such cellular transitions involve major or only minor modulation of chromatin structure. We have studied growth control and accessibility of chromatin to DNase I in C6 glioma cells subjected to different growth regimens using an in situ nick translation assay to label the most exposed regions of nuclear chromatin. In fibroblasts and primary glia, exposed chromatin was localized mainly at the nuclear lamina. This readily labeled DNA structure was largely lacking in the malignant C6 glioma. When C6 cells were treated with dibutyryl cyclic AMP, exposed chromatin was reestablished around the nuclear periphery. This restoration of a normal genome exposure pattern required cytoskeletal integrity. Thus large-scale nuclear reorganization events proceed in parallel with phenotypic normalization. The changes in cell morphology, growth control, cytoskeletal organization, and chromatin exposure and localization are similar to the reverse transformation reaction in CHO-K1 cells, which is also regulated by the cyclic nucleotide system. Hydrocortisone and dexamethasone also restored genome exposure in C6 but less markedly than cAMP derivatives. Diverse transformed cells can thus respond to growth control stimuli with similar nuclear restructuring events, which presumably underlie changes in gene expression. Reverse transformation and redifferentiation appear to be alternative terms describing essentially the same biological phenomenon.

Reverse transformation, genome exposure, and cancer.

The reverse transformation reaction whereby malignant cells are restored to a more normal phenotype has been reviewed. The primary causative action is ascribed to the genome exposure reaction in which a peripheral nuclear DNA region is restored to high sensitivity to DNase I, like that in normal cells. Various aspects of genome exposure around the nucleoli and the nuclear periphery are considered. The special role of the cytoskeleton in regulating exposure resulting in normal differentiation on the one hand and malignant transformation on the other is discussed. The action of the two-level system for regulation of the mammalian genome previously proposed is reviewed in relation to normal differentiation and malignancy with brief indication of roles played by various metabolites, transcription factors, protooncogenes, cell organelles, and processes like specific phosphorylation and dephosphorylation. Possible implications for cancer therapy and prevention and for the fields of genetic disease and toxicology are indicated.

Confocal microscopy of genome exposure in normal, cancer, and reverse-transformed cells.

Genome exposure studies were carried out on malignant CHO-K1 and C6 rat glioma cells and their respective, phenotypically normal counterparts (reverse-transformed CHO-K1, and both reverse-transformed C6 glioma and normal rat fibroblasts). Cells were subjected to the nick-translation technique previously developed to make visible the exposed (i.e., DNase I-sensitive) nuclear DNA, and examined by both epifluorescence and confocal microscopy. The confocal microscopy, by permitting examination of sections throughout the nucleus, made possible clearer identification of the regions of exposed and sequestered DNA in the cells studied. A peripheral shell of exposed DNA with some discontinuities was displayed in the great majority of the cells with normal phenotype, but in none of the cancer cells. Both types of cells displayed regions of exposed DNA in the nuclear interior, particularly surrounding the nucleoli. In accordance with previous theoretical proposals we postulate: the peripheral nuclear shell of exposed DNA contains differentiation-specific genes that include the specific growth-control genes and that are functional in normal cells but not in cancer; the exposed genes surrounding the nucleoli may represent housekeeping genes active in both normal and cancer cells; and the DNase I-resistant DNA in the interior of the nucleus we postulate to consist for the most part of genes specific to alternative differentiation states and to be sequestered and inactive. Previous differences in evaluation of roles of peripheral and internal DNA sensitivity to DNAse I hydrolysis appear to be reconciled by this formulation. Identification of exposed DNA may be useful in cancer diagnosis.

The spatial distribution of exposed nuclear DNA in normal, cancer, and reverse-transformed cells.

The malignant CHO-K1 cell is reverse-transformed by cAMP, regaining the phenotype of a normal fibroblast. During this reaction, much of its DNA re-acquires sensitivity to hydrolysis by DNase I in a way characteristic of the normal fibroblast. Exposed DNA forms a rim about the nucleus in both the normal and reverse-transformed cell but not in the malignant CHO-K1. Reacquisition of the nuclear rim requires an organized cytoskeleton. Sequestered DNA forms families of different degrees of sequestration. In accordance with previous theoretical developments it is proposed that (i) genes specific to a given differentiation state are stored in the nuclear rim, whereas genes specific to other states are sequestered within the nucleus; (ii) only exposed genes are active, and their activity is modulated by regulatory molecules in the fluid medium; (iii) exposure and sequestration are regulated by cytoskeletal and nuclear protein structures; (iv) in at least several types of cancer the regulatory defect lies in the genome exposure process so that the specific DNA sequences and their associated growth regulatory loci have been transferred from the exposed to the sequestered condition with consequent loss of the nuclear rim of exposed DNA. The methodology described should be generally applicable to examining the accessibility state of subsets of DNA during various physiological modulations of cell function.

Genome regulation in mammalian cells.

A theory is presented proposing that genetic regulation in mammalian cells is at least a two-tiered effect; that one level of regulation involves the transition between gene exposure and sequestration; that normal differentiation requires a different spectrum of genes to be exposed in each separate state of differentiation; that the fiber systems of the cell cytoskeleton and the nuclear matrix together control the degree of gene exposure; that specific phosphorylation of these elements causes them to assume a different organizational network and to impose a different pattern of sequestration and exposure on the elements of the genome; that the varied gene phosphorylation mechanisms in the cell are integrated in this function; that attachment of this network system to specific parts of the chromosomes brings about sequestration or exposure of the genes in their neighborhood in a fashion similar to that observed when microtubule elements attach through the kinetochore to the centromeric DNA; that one function of repetitive sequences is to serve as elements for the final attachment of this fibrous network to the specific chromosomal loci; and that at least an important part of the calcium manifestation as a metabolic trigger of different differentiation states involves its acting as a binding agent to centers of electronegativity, in particular proteins and especially phosphorylated groups, so as to change the conformation of the fiber network that ultimately controls gene exposure in the mammalian cell. It would appear essential to determine what abnormal gene exposures and sequestrations are characteristic of each type of cancer; which agonists, if any, will bring about reverse transformation; and whether these considerations can be used in therapy.

Conversion to the terminally differentiated state during treatment of NG108-15 neural tumor cells with 1-beta-D-arabinofuranosylcytosine in defined medium.

Differentiation of cells of the neural tumor hybrid cloned cell line NG108-15 begins stochastically upon transfer to a serum-free defined medium (A. Krystosek, J. Cell. Physiol., 125: 319-329, 1985). Such cultures in N2 medium contain both proliferating and neurite-forming cells (which are not mutally exclusive subpopulations). Addition of 1-beta-D-arabinofuranosylcytosine (ara-C) to NG108-15 cells in N2 medium yielded cultures with a highly differentiated appearance. Investigation of the nature of this effect revealed that ara-C did not increase the probability that cells would enter the differentiation pathway; it did, however, completely abolish proliferation. The early kinetics of neurite formation were similar in control and treated cultures. This was followed by a phase in which ara-C-treated cells underwent continuous rapid maturation including normalization of nucleolar features. Loss of proliferative potential of treated cells was tested in a drug-free serum challenge protocol. Permanently postmitotic cells (i.e., cells which failed to divide even once) were shown to accumulate with time of ara-C pretreatment; this represented 59% of total cells at day 3 and 94% at day 7 of treatment. Thus, the bulk of the population can commit to terminal differentiation. Even among the minority of cells capable of 1-2 rounds of division in the challenge incubation, cessation of proliferation was more likely than continuous colonial growth, suggesting that a profound phenotypic alteration had occurred. The results show that advanced morphological maturation and the step(s) of terminal neuronal differentiation can be achieved in this cell line in response to a cancer chemotherapeutic agent and that this drug is a more complete inducer than compounds which modulate the cyclic AMP system.

Plasminogen activator secretion in relation to Schwann cell activities.

The molecular characterization of neural secretory plasminogen activators and their possible association with two well-defined Schwann cell activities (mitosis and migration) were investigated. Schwann cells from peripheral sensory ganglion roots were established as primary cultures of high purity. Conditioned culture medium was fractionated by gel electrophoresis and assayed for plasminogen-dependent in situ caseinolytic activity. Both tissue-type and urokinase-type plasminogen activators were detected early in culture. The amount of the tissue-type form decreased with increasing time in culture. Proliferative ability measured as [3H]thymidine incorporation into the nucleus varied under several culture conditions and was lowest in defined serum-free medium; however, these serum-free cultures expressed the highest level of plasminogen activator secreting cells. Combined autoradiography/fibrin overlay assays allowed for a direct analysis at the single cell level; interestingly the findings did not support the idea of an obligatory expression of the enzyme in replicating cells. Cell motility was assayed in combined gold particle clearing/fibrin overlay assays and likewise no obligatory association of protease secretion and cell movement could be demonstrated. The results are discussed in relation to previous studies in this field.

Normal and malignant cells, including neurons, deposit plasminogen activator on the growth substrata.

The results of four different assay methods showed that both normal and malignant plasminogen activator-secreting cells deposited substantial amounts of this protease on tissue-culture substrata, including collagen coatings. The cells studied were Rous sarcoma virus (RSV)-transformed vole fibroblasts, a malignant neural cell line (NG108-15) capable of neurite formation, and normal mouse-regenerating sensory neurons. Deposited plasminogen activator was detected by a fibrin overlay assay at sites from which cells growing on coverslips had been gently dislodged, showing that active enzyme is left beneath cells and in the immediate pericellular area. For neuronal cells, fibrinolytic zones were detected not only at the previous positions of cell bodies but also along the terrain conditioned by neurite extension, suggesting that a trail of plasminogen activator is left behind during growth cone movement. Substratum-bound enzyme could be solubilized in buffers containing sodium dodecyl sulfate (SDS) or Triton X-100 and demonstrated by zymography following electrophoresis or assayed for amidolytic activity with a chromogenic substrate (Kabi S-2251). The results suggest that plasminogen activator may be considered a component of substrate-adhesion material. Secretory proteases deposited directly on matrix molecules would seem strategically positioned to participate in local degradation of components of the extracellular environment.

Neurite formation by neuroblastoma-glioma hybrid cells (NG108-15) in defined medium: stochastic initiation with persistence of differentiated functions.

Neurite formation and proliferation by NG108-15 cells were studied in short-term serum-free N2 medium. Neuritogenesis by individual cells was observed at widely differing times, suggesting a stochastic component to initiation of differentiation. Cells with and without neurites could also enter one or more rounds of proliferation at varying times. The initial choice between these divergent behaviors influenced subsequent growth. Cells initially extending neurites had a high probability of continuing neuritic elongation. Cells initially dividing had a high probability of yielding further progeny. Addition of dibutyryl cyclic AMP to cells cultured in N2 medium rapidly increased the probability of differentiating and decreased the probability of proliferating. To test whether or not cells with highly differentiated morphologies had irreversibly lost the capacity for proliferation, induced cultures were washed and challenged by the addition of serum-containing medium. The length of time required for individual cells to divide increased with increasing time of preincubation in the induction medium. However, few cells appeared to be permanently removed from the proliferative pool. These observations suggest that differentiating cells exhibit persistence, a tendency to continue on the differentiation pathway. Persistence is extinguished following one round of proliferation in serum-containing medium.

Peripheral neurons and Schwann cells secrete plasminogen activator.

The secretion of the protease plasminogen activator (PA) by cells of developing peripheral nerve was demonstrated. Fetal and early postnatal dorsal root ganglia were established in culture as explants or as individual neurons and Schwann cells. A fibrin overlay assay was used to visualize the locations of PA secretion. Fibrinolytic zones formed around the somata of explants and were skewed in the direction of maximal fiber outgrowth. Individual growth cones at the tips of long fasiculated fiber bundles also released PA. Approximately 50% of individual neurons showed PA secretion; especially pronounced release occurred at some growth cones. Culture of nerve growth factor-independent adult neurons showed that PA expression was independent of effects of this growth hormone. A subpopulation of Schwann cells was also active in PA secretion, which could be detected at the soma, at the bipolar processes, or along the entire cell length. Possible functions of neural PA in development and regeneration are discussed.

Plasminogen activator secretion by granule neurons in cultures of developing cerebellum.

Dispersed cell cultures were established from 7- to 9-day postnatal mouse cerebellum. The fibrin slice method was used to obtain a localization of plasminogen activator production to specific cells. Fibrinolytically active cells were small (5- to 8-micrometer diameter), round, and occurred singly or in aggregates. Fibrinolysis was both plasminogen and time dependent, inhibitable by epsilon-aminocaproic acid and soybean trypsin inhibitor and did not occur when cells were fixed in formalin prior to the fibrin overlay. Strong fibrin degradation occurred only when granule neurons were abundant in the cultures. These plasminogen activator secreting cells were identified as granule neurons by cell separation methods, nuclear morphology, and their ability to bind tetanus toxin and rabbit antiserum against mouse cerebellum (anti-Cbl-1 antiserum). Plasminogen activator also could be quantified in fractionated tissue homogenates or in cell culture medium by the 125I-labeled fibrin plate assay. Fibrinolysis in cerebellar extracts was 95% dependent on the presence of added plasminogen; furthermore, the activity was greater in cerebellar extracts as compared to cerebral cortex of the same age. At the age examined, the cerebellum contains many migratory neurons, and plasminogen activator production may be involved in the process of cell movement.

Plasminogen activator release at the neuronal growth cone.

The site of plasminogen activator release by differentiated neuroblastoma clonal cell lines was determined with a fibrin overlay assay. Release of plasminogen activator was seen at the growth cone in 72 percent of the cells bearing neurites. For 21 percent of these cells the growth cone was the predominant or exclusive site of this enzyme activity. Selective release of protease at the "trailblazing" tip of the neurite may be important in neuron migration and neurite growth in vivo.

Control of lysozyme induction in the differentiation of myeloid leukemic cells.

A system has been developed with clones of mouse myeloid leukemic cells in culture to study the induction of synthesis and secretion of lysozyme in relation to other steps in myeloid cell differentiation. Lysozyme was initially absent in all the clones studied. The different clones can be divided into three types according to their ability to be induced to undergo normal cell differentiation by the protein inducer MGI (macrophage and granulocyte inducer). One type of clone that can be induced by MGI to form Fc and C3 receptors and differentiate to mature macrophages and granulocytes (MGI+D+) was also induced by MGI to synthesize and secrete lysozyme. Lysozyme was induced after Fc and C3 receptors, and labeling with 35S-methionine has shown that the induced lysozyme was newly synthesized. MGI+D+ clones were also induced to synthesize and secrete lysozyme by dexamethasone, prednisolone, cytosine arabinoside, or thymidine and in one of four clones by dimethylsulfoxide but not by sodium butyrate. Inhibition of cell multiplication by itself was not sufficient to induce lysozyme synthesis. A second type of clone which can be induced by MGI to form Fc and C3 receptors but not mature cells (MGI+D-) was more weakly inducible by MGI for lysozyme and was not inducible by any of the other compounds. A third type of clone (MGI-D) MGI for receptors or mature cells. One of four MGI-D- clones was induced to synthesize but not secrete lysozyme by dexamethasone together with sodium butyrate, but there was no lysozyme induction by MGI or any of the other compounds separately. The different clones maintained their different properties for at least 6 months in culture. The results indicate that clones with different hereditary blocks in the ability to be induced to differentiate by specific compounds can be used to dissect the process of myeloid cell differentiation, that the sequence of differentiation is induction of Fc and C3 receptors leads to lysozyme leads to mature cells, and that there are separate controls for these developmental steps.

Improved methods for purification and assay of eukaryotic messenger ribonucleic acids and ribosomes. Quantitative analysis of their interaction in a fractionated reticulocyte cell-free system.

The polyadenylic acid-containing messenger ribonucleic acids of eukaryotic cells are rapidly isolated and deproteinized in a simple and gentle one-step procedure. The polyribosome fraction, dissolved in 0.5 M NaCl/0.5 percent sodium dodecyl sulfate, is passed through an oligo(dT)-cellulose column which is then washed with the solvent until proteins and contaminating ribonucleic acids are fully removed. Deproteinized messenger ribonucleic acid is then eluted by lowering the ionic strength. This method gives highly purified and active messenger ribonucleic acids from all tissues tested. The yield is approximately 1.5 to 2 percent of the polyribosomal ribonucleic acid. Messenger ribonucleic acids are assayed in a rabbit reticulocyte-derived, messenger-dependent, cell-free protein-synthesizing system modified from Crystal et al. (Crystal, R. G., Nienhuis, A. W., Elson, N. A., and Anderson, W.F. (1972) J. Biol. Chem. 247, 5357-5368). This system synthesizes proteins at an almost linear rate for at least 2 hours. During this period, each globin messenger ribonucleic acid directs the synthesis of several globin molecules. Each active ribosome synthesizes a globin molecule every 6 to 7 min, but only a small fraction of the ribosomes or messengers are active at any instant. Translation occurs mainly on di- and monoribosomes although larger sized polysomes also occur. Several lines of evidence suggest that globin messenger ribonucleic acid requires "activation" before it can be utilized and that a messenger activation step of protein synthesis initiation is rate-limiting in this cell-free system.